<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: Bhavik Pahwa</title>
    <description>The latest articles on DEV Community by Bhavik Pahwa (@cren).</description>
    <link>https://dev.to/cren</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4162245%2Fd8c82877-9784-4b50-8d05-c4df45caaf01.png</url>
      <title>DEV Community: Bhavik Pahwa</title>
      <link>https://dev.to/cren</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/cren"/>
    <language>en</language>
    <item>
      <title>Funny Songs 101</title>
      <dc:creator>Bhavik Pahwa</dc:creator>
      <pubDate>Thu, 22 Oct 2026 16:37:21 +0000</pubDate>
      <link>https://dev.to/cren/funny-songs-101-398o</link>
      <guid>https://dev.to/cren/funny-songs-101-398o</guid>
      <description>&lt;p&gt;&lt;em&gt;This is a submission for the &lt;a href="https://dev.to/challenges/hacktoberfest-week3-2026-10-19"&gt;Hacktoberfest Open-Source AI Challenge Week 3: Slow and Steady&lt;/a&gt;&lt;/em&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  What I Built
&lt;/h2&gt;

&lt;p&gt;I built &lt;strong&gt;The Lazy Jukebox&lt;/strong&gt; (aka &lt;em&gt;Endless Sloth FM&lt;/em&gt;). I often get bored of my own playlists, but finding new, entertaining music takes too much effort. &lt;/p&gt;

&lt;p&gt;The Lazy Jukebox is a completely offline CLI application that endlesssly generates and plays funny, spoken-word "songs" and raps. You run the script once, sit back, and the machine handles all the creative thinking, lyric writing, vocal synthesis, and DJ mixing indefinitely. It saves you from the dreadful task of curating playlists while rewarding maximum laziness with infinite laughs.&lt;/p&gt;

&lt;h2&gt;
  
  
  Demo
&lt;/h2&gt;

&lt;p&gt;  &lt;iframe src="https://www.youtube.com/embed/jzm8l0gIdbU" width="710" height="399"&gt;
  &lt;/iframe&gt;
&lt;/p&gt;

&lt;h2&gt;
  
  
  Code
&lt;/h2&gt;


&lt;div class="ltag-github-readme-tag"&gt;
  &lt;div class="readme-overview"&gt;
    &lt;h2&gt;
      &lt;img src="https://assets.dev.to/assets/github-logo-5a155e1f9a670af7944dd5e12375bc76ed542ea80224905ecaf878b9157cdefc.svg" alt="GitHub logo"&gt;
      &lt;a href="https://github.com/Bhavik-Pahwa" rel="noopener noreferrer"&gt;
        Bhavik-Pahwa
      &lt;/a&gt; / &lt;a href="https://github.com/Bhavik-Pahwa/Funny-Songs-101" rel="noopener noreferrer"&gt;
        Funny-Songs-101
      &lt;/a&gt;
    &lt;/h2&gt;
    &lt;h3&gt;
      
    &lt;/h3&gt;
  &lt;/div&gt;
  &lt;div class="ltag-github-body"&gt;
    
&lt;div id="readme" class="md"&gt;&lt;div class="markdown-heading"&gt;
&lt;h1 class="heading-element"&gt;The Lazy Jukebox / Endless Sloth FM&lt;/h1&gt;
&lt;/div&gt;
&lt;p&gt;Run once, sit back, and let a local AI DJ invent ridiculous spoken-word songs
Wonderwords collides a noun, adjective, and everyday predicament; Ollama writes
the lyrics; Piper speaks them; pydub mixes a beat; pygame plays the result while
the next track is being made. No cloud API keys, subscriptions, or runtime downloads.&lt;/p&gt;
&lt;div class="markdown-heading"&gt;
&lt;h2 class="heading-element"&gt;Setup&lt;/h2&gt;
&lt;/div&gt;
&lt;p&gt;Use &lt;strong&gt;64-bit Python 3.10–3.13&lt;/strong&gt;. Python 3.13 is tested; Python 3.14 is not yet a
supported target because audio dependencies may lack compatible wheels.
Install &lt;a href="https://ollama.com/download" rel="nofollow noopener noreferrer"&gt;Ollama&lt;/a&gt; and &lt;a href="https://ffmpeg.org/download.html" rel="nofollow noopener noreferrer"&gt;FFmpeg&lt;/a&gt;
(both &lt;code&gt;ffmpeg&lt;/code&gt; and &lt;code&gt;ffprobe&lt;/code&gt; must be on PATH).&lt;/p&gt;
&lt;p&gt;From this directory, on Windows PowerShell:&lt;/p&gt;
&lt;div class="highlight highlight-source-powershell notranslate position-relative overflow-auto js-code-highlight"&gt;
&lt;pre&gt;py &lt;span class="pl-c1"&gt;-3.13&lt;/span&gt; &lt;span class="pl-k"&gt;-&lt;/span&gt;m venv .venv
.\.venv\Scripts\Activate.ps1
python &lt;span class="pl-k"&gt;-&lt;/span&gt;m pip install &lt;span class="pl-k"&gt;-&lt;/span&gt;r requirements.txt
ollama pull llama3.&lt;span class="pl-c1"&gt;2&lt;/span&gt;:1b
python &lt;span class="pl-k"&gt;-&lt;/span&gt;m piper.download_voices en_US&lt;span class="pl-k"&gt;-&lt;/span&gt;lessac&lt;span class="pl-k"&gt;-&lt;/span&gt;medium &lt;span class="pl-k"&gt;--&lt;/span&gt;data&lt;span class="pl-k"&gt;-&lt;/span&gt;dir voices&lt;/pre&gt;

&lt;/div&gt;
&lt;p&gt;On macOS/Linux:&lt;/p&gt;
&lt;div class="highlight highlight-source-shell notranslate position-relative overflow-auto js-code-highlight"&gt;
&lt;pre&gt;python3 -m venv .venv
&lt;span class="pl-c1"&gt;source&lt;/span&gt; .venv/bin/activate
python -m pip install -r requirements.txt&lt;/pre&gt;…
&lt;/div&gt;&lt;/div&gt;
  &lt;/div&gt;
  &lt;div class="gh-btn-container"&gt;&lt;a class="gh-btn" href="https://github.com/Bhavik-Pahwa/Funny-Songs-101" rel="noopener noreferrer"&gt;View on GitHub&lt;/a&gt;&lt;/div&gt;
&lt;/div&gt;


&lt;h2&gt;
  
  
  How I Built It
&lt;/h2&gt;

&lt;p&gt;I wanted this to run locally on lower-spec hardware (8GB RAM, integrated graphics), which meant heavy music generation models were out. Instead, I used a clever combination of lightweight open-source AI tools:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;Idea Engine:&lt;/strong&gt; A Python "Concept Collider" using &lt;code&gt;wonderwords&lt;/code&gt; pairs random adjectives/nouns with mundane activities (e.g., "A rap about a radioactive cabbage doing taxes").&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Lyric Generation:&lt;/strong&gt; &lt;strong&gt;Ollama&lt;/strong&gt; running locally with the lightweight &lt;strong&gt;Llama 3.2 (1B)&lt;/strong&gt; model. It takes the prompt and instantly writes a 4-stanza comedy rap. &lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Vocal Synthesis:&lt;/strong&gt; &lt;strong&gt;Piper TTS&lt;/strong&gt;, an insanely fast and low-resource neural text-to-speech engine optimized for CPUs. Using a highly expressive voice model, it speaks the lyrics like deadpan slam poetry.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;The DJ:&lt;/strong&gt; Python's &lt;code&gt;pydub&lt;/code&gt; mixes the generated vocals over local royalty-free &lt;code&gt;.mp3&lt;/code&gt; hip-hop beats, and &lt;code&gt;pygame&lt;/code&gt; handles the endless playback loop. While one song plays, the next generates in the background!&lt;/li&gt;
&lt;/ol&gt;

&lt;h2&gt;
  
  
  Why Does Open Innovation Matter?
&lt;/h2&gt;

&lt;p&gt;Open innovation is the &lt;em&gt;only&lt;/em&gt; reason this project is possible. If I built an "endless" loop using closed-source APIs like OpenAI for text and ElevenLabs for voice, a few hours of laziness would bankrupt me in API costs! &lt;/p&gt;

&lt;p&gt;By using open-weight models via Ollama and open-source local inference via Piper TTS, my app runs entirely offline, costs $0 to operate forever, and keeps my data completely private. Open AI makes infinite laziness affordable.&lt;/p&gt;

&lt;h2&gt;
  
  
  The Lazy Payoff
&lt;/h2&gt;

&lt;p&gt;This app saves me the 20 minutes I usually waste skipping tracks on Spotify trying to find something fun. I ran it this weekend while cleaning my room, and the AI DJ entertained me for two hours straight without me ever touching the keyboard. Pure sloth perfection. &lt;/p&gt;

&lt;h2&gt;
  
  
  Prize Categories
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Open-Source AI&lt;/li&gt;
&lt;li&gt;Local AI Models&lt;/li&gt;
&lt;/ul&gt;

</description>
      <category>devchallenge</category>
      <category>hf26challenge</category>
    </item>
    <item>
      <title>Your Content Help Buddy</title>
      <dc:creator>Bhavik Pahwa</dc:creator>
      <pubDate>Fri, 16 Oct 2026 18:32:05 +0000</pubDate>
      <link>https://dev.to/cren/your-content-help-buddy-38h2</link>
      <guid>https://dev.to/cren/your-content-help-buddy-38h2</guid>
      <description>&lt;p&gt;&lt;em&gt;This is a submission for the &lt;a href="https://dev.to/challenges/hacktoberfest-week2-2026-10-12"&gt;Hacktoberfest Open-Source AI Challenge Week 2: Side Hustle&lt;/a&gt;&lt;/em&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  What I Built
&lt;/h2&gt;

&lt;p&gt;I built &lt;strong&gt;Reelflow&lt;/strong&gt;, a content ideation tool that helps creators turn fresh stories in their niche into a rough Instagram Reel script.&lt;/p&gt;

&lt;p&gt;The problem starts before someone presses record: finding something worth talking about, understanding the story, and turning it into a short video with a clear beginning and ending. For a solo creator or freelance social media manager, repeating that research for every Reel can take time away from actually creating.&lt;/p&gt;

&lt;p&gt;Reelflow brings those steps into one focused flow:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;Choose a niche:&lt;/strong&gt; AI, technology, startups, coding, design, or space.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Find a spark:&lt;/strong&gt; explore cards containing fresh stories, source links, publication dates, and available community signals.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Get the context:&lt;/strong&gt; review the source highlights and choose a 30-, 45-, or 60-second Reel with a conversational, educational, or bold delivery.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Generate the flow:&lt;/strong&gt; get a &lt;strong&gt;hook, plot, twist, and conclusion&lt;/strong&gt;, with voiceover text, shot suggestions, and on-screen overlays.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Make it yours:&lt;/strong&gt; edit the voiceover, then copy or download the script with its source attribution.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;There is no login or signup. The interface moves through one stage at a time, using warm colors, modern typography, and animated transitions to make the process feel like a creative onboarding experience.&lt;/p&gt;

&lt;p&gt;The potential paying customer is a creator or freelance social media manager who needs to research and outline niche content regularly. The value I want to sell is less time between finding a story and having a usable recording plan. Willingness to pay is still a hypothesis; I have not validated it with paying customers.&lt;/p&gt;

&lt;h2&gt;
  
  
  Demo
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;Demo video:&lt;/strong&gt;   &lt;iframe src="https://www.youtube.com/embed/ZShTy-lQwKE" width="710" height="399"&gt;
  &lt;/iframe&gt;
&lt;/p&gt;

&lt;p&gt;The current version runs locally. It does not have a public deployment yet. To try it with Node.js 22+ and Ollama installed, run:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;ollama pull qwen3:8b
npm start
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;


&lt;p&gt;Then open &lt;code&gt;http://localhost:3000&lt;/code&gt; with Ollama running.&lt;/p&gt;

&lt;p&gt;A walkthrough can follow the complete journey: select Space &amp;amp; Discovery, choose a live article, review its summary, generate a 30-second Reel, edit a line, and download the finished draft.&lt;/p&gt;

&lt;p&gt;During development, all six source feeds returned live stories. A real generation using Qwen3 8B produced all four script stages in approximately &lt;strong&gt;96 seconds on the development machine's CPU&lt;/strong&gt;. That is one observed run, not a guaranteed generation time. Browser visual verification was not completed in the development session.&lt;/p&gt;
&lt;h2&gt;
  
  
  Code
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;GitHub repository:&lt;/strong&gt; &lt;/p&gt;
&lt;div class="ltag-github-readme-tag"&gt;
  &lt;div class="readme-overview"&gt;
    &lt;h2&gt;
      &lt;img src="https://assets.dev.to/assets/github-logo-5a155e1f9a670af7944dd5e12375bc76ed542ea80224905ecaf878b9157cdefc.svg" alt="GitHub logo"&gt;
      &lt;a href="https://github.com/Bhavik-Pahwa" rel="noopener noreferrer"&gt;
        Bhavik-Pahwa
      &lt;/a&gt; / &lt;a href="https://github.com/Bhavik-Pahwa/Content-Ideation-Tool" rel="noopener noreferrer"&gt;
        Content-Ideation-Tool
      &lt;/a&gt;
    &lt;/h2&gt;
    &lt;h3&gt;
      
    &lt;/h3&gt;
  &lt;/div&gt;
  &lt;div class="ltag-github-body"&gt;
    
&lt;div id="readme" class="md"&gt;&lt;div class="markdown-heading"&gt;
&lt;h1 class="heading-element"&gt;Reelflow&lt;/h1&gt;
&lt;/div&gt;

&lt;p&gt;A local, no-login content studio that turns fresh niche stories into editable Instagram reel drafts. Six niches, live source cards, source highlights, and four scripted scenes: &lt;strong&gt;Hook → Plot → Twist → Conclusion&lt;/strong&gt;.&lt;/p&gt;

&lt;div class="markdown-heading"&gt;
&lt;h2 class="heading-element"&gt;Run&lt;/h2&gt;
&lt;/div&gt;

&lt;p&gt;Requirements: Node.js 22+ and &lt;a href="https://ollama.com/" rel="nofollow noopener noreferrer"&gt;Ollama&lt;/a&gt;. No npm dependencies or API keys are required.&lt;/p&gt;

&lt;div class="highlight highlight-source-powershell notranslate position-relative overflow-auto js-code-highlight"&gt;
&lt;pre&gt;ollama pull qwen3:8b
npm start&lt;/pre&gt;

&lt;/div&gt;
&lt;p&gt;Open &lt;a href="http://localhost:3000" rel="nofollow noopener noreferrer"&gt;http://localhost:3000&lt;/a&gt;. Ollama must be running. On this machine, Qwen3 8B was already installed. A first request may take longer while the model loads. Scripts have a four-minute timeout, with a visible error and retry path. Run &lt;code&gt;npm test&lt;/code&gt; for server contract tests.&lt;/p&gt;
&lt;p&gt;Optional PowerShell configuration:&lt;/p&gt;
&lt;div class="highlight highlight-source-powershell notranslate position-relative overflow-auto js-code-highlight"&gt;
&lt;pre&gt;&lt;span class="pl-smi"&gt;$&lt;span class="pl-c1"&gt;env:&lt;/span&gt;OLLAMA_MODEL&lt;/span&gt; &lt;span class="pl-k"&gt;=&lt;/span&gt; &lt;span class="pl-s"&gt;&lt;span class="pl-pds"&gt;'&lt;/span&gt;qwen3:8b&lt;span class="pl-pds"&gt;'&lt;/span&gt;&lt;/span&gt;
&lt;span class="pl-smi"&gt;$&lt;span class="pl-c1"&gt;env:&lt;/span&gt;OLLAMA_URL&lt;/span&gt; &lt;span class="pl-k"&gt;=&lt;/span&gt; &lt;span class="pl-s"&gt;&lt;span class="pl-pds"&gt;'&lt;/span&gt;http://127.0.0.1:11434&lt;span class="pl-pds"&gt;'&lt;/span&gt;&lt;/span&gt;
&lt;span class="pl-smi"&gt;$&lt;span class="pl-c1"&gt;env:&lt;/span&gt;PORT&lt;/span&gt; &lt;span class="pl-k"&gt;=&lt;/span&gt; &lt;span class="pl-s"&gt;&lt;span class="pl-pds"&gt;'&lt;/span&gt;3000&lt;span class="pl-pds"&gt;'&lt;/span&gt;&lt;/span&gt;
npm start&lt;/pre&gt;

&lt;/div&gt;
&lt;p&gt;The app listens on loopback intentionally. It is a local MVP, not a multi-tenant production service. A public deployment needs an inference host, request quotas…&lt;/p&gt;&lt;/div&gt;
  &lt;/div&gt;
  &lt;div class="gh-btn-container"&gt;&lt;a class="gh-btn" href="https://github.com/Bhavik-Pahwa/Content-Ideation-Tool" rel="noopener noreferrer"&gt;View on GitHub&lt;/a&gt;&lt;/div&gt;
&lt;/div&gt;

&lt;p&gt;&lt;br&gt;&lt;br&gt;
&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fo3fn7y5le4zh80b8xjpl.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fo3fn7y5le4zh80b8xjpl.png" alt=" " width="800" height="425"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fdrdpk2zox71nifiaoblo.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fdrdpk2zox71nifiaoblo.png" alt=" " width="800" height="425"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;The project includes an MIT license, setup instructions, API research notes, and five passing server tests.&lt;/p&gt;

&lt;p&gt;The main files are:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;code&gt;server.mjs&lt;/code&gt;: source integrations, feed caching, Ollama requests, and generated-output validation.&lt;/li&gt;
&lt;li&gt;
&lt;code&gt;public/app.js&lt;/code&gt;: the niche-to-script journey, editable voiceovers, and exports.&lt;/li&gt;
&lt;li&gt;
&lt;code&gt;public/style.css&lt;/code&gt;: responsive layouts, transitions, and reduced-motion support.&lt;/li&gt;
&lt;li&gt;
&lt;code&gt;server.test.mjs&lt;/code&gt;: server contracts, generation input checks, and malformed-output checks.&lt;/li&gt;
&lt;li&gt;
&lt;code&gt;docs/API-RESEARCH.md&lt;/code&gt;: the niche-to-API mapping and source limitations.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  How I Built It
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;Qwen3 8B, an open-weight model running locally through Ollama, is the creative engine.&lt;/strong&gt; Every generated script comes from the model; the app does not substitute a canned script when inference fails.&lt;/p&gt;

&lt;p&gt;The application uses a Node.js HTTP server and a plain HTML, CSS, and JavaScript frontend. It has no npm runtime dependencies. The backend fetches stories from free public API endpoints and caches each niche's feed for five minutes:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;a href="https://hn.algolia.com/api" rel="noopener noreferrer"&gt;Hacker News search through Algolia&lt;/a&gt; supplies recent AI, technology, and startup stories with points and comment counts.&lt;/li&gt;
&lt;li&gt;
&lt;a href="https://developers.forem.com/api/v1" rel="noopener noreferrer"&gt;The DEV API&lt;/a&gt; supplies coding and design posts with descriptions and community reactions.&lt;/li&gt;
&lt;li&gt;
&lt;a href="https://www.spaceflightnewsapi.net/" rel="noopener noreferrer"&gt;Spaceflight News API&lt;/a&gt; supplies recent space reporting with publisher summaries.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;These signals need context: community activity is not proof that something is trending on Instagram, and recent space reporting is not a popularity ranking. Reelflow labels the distinction instead of inventing a viral score.&lt;/p&gt;

&lt;p&gt;After a user selects a story, the server sends its available source context, desired duration, and delivery style to Ollama's &lt;code&gt;/api/chat&lt;/code&gt; endpoint. A JSON schema defines the four required scenes and their voiceover, visual, and overlay fields. The server validates the response again before displaying it.&lt;/p&gt;

&lt;p&gt;The prompt tells the model to treat source text as data, use only supplied facts, and avoid invented statistics or quotes. When a source supplies only a headline, the interface discloses that limitation and the prompt requests an exploratory angle. These safeguards do not guarantee factual accuracy: the creator still needs to check the original source before recording.&lt;/p&gt;

&lt;p&gt;I used Codex to help implement and test the application. The generated Reel content itself is powered by the local Qwen model.&lt;/p&gt;

&lt;h2&gt;
  
  
  Why Does Open Innovation Matter?
&lt;/h2&gt;

&lt;p&gt;Open innovation makes the economics and ownership of this tool more flexible.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Experimenting does not incur a paid inference API charge for each draft.&lt;/strong&gt; A creator can change the tone or regenerate an outline using their own machine. Hardware and electricity still cost money, and running a hosted service would add costs, but there is no per-generation model-provider bill in the local setup.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;The writing process stays on the machine running Reelflow.&lt;/strong&gt; With the default local configuration, the selected story context and generated script are processed by Ollama on that computer. Discovering fresh stories still requires internet access, so this is not a fully offline research tool.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;The creative engine can change with the product.&lt;/strong&gt; The model is configurable through &lt;code&gt;OLLAMA_MODEL&lt;/code&gt;, and the prompt and output schema are directly editable. I can adapt the workflow to a specific creator's niche or try another compatible local model without rebuilding around a different paid API contract. This version uses the existing Qwen3 model; I have not fine-tuned it.&lt;/p&gt;

&lt;p&gt;A closed API could also generate a script. The difference here is that I can run and inspect the application around a model I control locally, choose when to change it, and keep inference independent of a hosted provider's pricing. For a small tool aimed at repeat creative work, those choices matter.&lt;/p&gt;

</description>
      <category>devchallenge</category>
      <category>hf26challenge</category>
      <category>ai</category>
      <category>webdev</category>
    </item>
    <item>
      <title>The Guilt-Trip Potato</title>
      <dc:creator>Bhavik Pahwa</dc:creator>
      <pubDate>Thu, 08 Oct 2026 09:49:27 +0000</pubDate>
      <link>https://dev.to/cren/the-guilt-trip-potato-2j9a</link>
      <guid>https://dev.to/cren/the-guilt-trip-potato-2j9a</guid>
      <description>&lt;p&gt;&lt;em&gt;This is a submission for the &lt;a href="https://dev.to/challenges/hacktoberfest-week1-2026-10-05"&gt;Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass&lt;/a&gt;&lt;/em&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  What I Built
&lt;/h2&gt;

&lt;p&gt;I built SpudBot, a heavily sarcastic, locally-hosted desktop assistant that aggressively ensures you step away from your screen.&lt;/p&gt;

&lt;p&gt;SpudBot starts as a tiny, cutesy potato icon sitting quietly on your desktop. But if you spend too much uninterrupted time staring at your laptop, SpudBot gets angry. It enlarges to 5x its original size and starts relentlessly guilt-tripping you with unique, AI-generated taunts about your lack of vitamin D and social life.&lt;/p&gt;

&lt;p&gt;If you ignore the taunts and refuse to get up, SpudBot escalates. It will start messing with your Windows environment—randomly dimming your screen brightness, locking your touchpad, or straight-up closing a random window. It’s built for any developer who loses track of time and needs a (literally) disruptive push to close the lid and finally touch grass.&lt;/p&gt;

&lt;h2&gt;
  
  
  Demo
&lt;/h2&gt;

&lt;p&gt;  &lt;iframe src="https://www.youtube.com/embed/fA2FdBuoJb8" width="710" height="399"&gt;
  &lt;/iframe&gt;
&lt;/p&gt;

&lt;h2&gt;
  
  
  Code
&lt;/h2&gt;


&lt;div class="ltag-github-readme-tag"&gt;
  &lt;div class="readme-overview"&gt;
    &lt;h2&gt;
      &lt;img src="https://assets.dev.to/assets/github-logo-5a155e1f9a670af7944dd5e12375bc76ed542ea80224905ecaf878b9157cdefc.svg" alt="GitHub logo"&gt;
      &lt;a href="https://github.com/Bhavik-Pahwa" rel="noopener noreferrer"&gt;
        Bhavik-Pahwa
      &lt;/a&gt; / &lt;a href="https://github.com/Bhavik-Pahwa/Guilt-Trip-Potato" rel="noopener noreferrer"&gt;
        Guilt-Trip-Potato
      &lt;/a&gt;
    &lt;/h2&gt;
    &lt;h3&gt;
      
    &lt;/h3&gt;
  &lt;/div&gt;
  &lt;div class="ltag-github-body"&gt;
    
&lt;div id="readme" class="md"&gt;&lt;div class="markdown-heading"&gt;
&lt;h1 class="heading-element"&gt;SpudBot · the guilt-trip potato&lt;/h1&gt;
&lt;/div&gt;
&lt;p&gt;A small desktop companion with increasingly large opinions about your screen time
Built for Windows with Python 3.10+ and Tkinter. All three placeholder potato sprites
are drawn in code: no image downloads, icon packages, accounts, or API keys needed.&lt;/p&gt;
&lt;div class="markdown-heading"&gt;
&lt;h2 class="heading-element"&gt;Run&lt;/h2&gt;
&lt;/div&gt;
&lt;p&gt;Install Python from &lt;a href="https://www.python.org/downloads/windows/" rel="nofollow noopener noreferrer"&gt;python.org&lt;/a&gt; with
Tcl/Tk support (included in the regular installer). In this directory:&lt;/p&gt;
&lt;div class="highlight highlight-source-powershell notranslate position-relative overflow-auto js-code-highlight"&gt;
&lt;pre&gt;python &lt;span class="pl-k"&gt;-&lt;/span&gt;m spudbot&lt;/pre&gt;

&lt;/div&gt;
&lt;p&gt;Or double-click &lt;code&gt;launch.bat&lt;/code&gt;. No Python packages are required for the core app.&lt;/p&gt;
&lt;div class="highlight highlight-source-powershell notranslate position-relative overflow-auto js-code-highlight"&gt;
&lt;pre&gt;python &lt;span class="pl-k"&gt;-&lt;/span&gt;m spudbot &lt;span class="pl-k"&gt;--&lt;/span&gt;demo&lt;/pre&gt;

&lt;/div&gt;
&lt;p&gt;Demo runs at 60× speed: default moods change after 30 and 60 real seconds.
Every OS effect is simulated, even if enabled in preferences. Manual breaks still
use real minutes. Switching demo on/off starts a fresh session.&lt;/p&gt;
&lt;div class="markdown-heading"&gt;
&lt;h2 class="heading-element"&gt;Local AI&lt;/h2&gt;

&lt;/div&gt;
&lt;p&gt;Install &lt;a href="https://ollama.com/download/windows" rel="nofollow noopener noreferrer"&gt;Ollama&lt;/a&gt;, then download a model once:&lt;/p&gt;
&lt;div class="highlight highlight-source-powershell notranslate position-relative overflow-auto js-code-highlight"&gt;
&lt;pre&gt;ollama pull llama3.&lt;span class="pl-c1"&gt;2&lt;/span&gt;:1b&lt;/pre&gt;

&lt;/div&gt;
&lt;p&gt;Start the Ollama desktop application (or &lt;code&gt;ollama serve&lt;/code&gt; if it isn't already running)
Choose the…&lt;/p&gt;&lt;/div&gt;
  &lt;/div&gt;
  &lt;div class="gh-btn-container"&gt;&lt;a class="gh-btn" href="https://github.com/Bhavik-Pahwa/Guilt-Trip-Potato" rel="noopener noreferrer"&gt;View on GitHub&lt;/a&gt;&lt;/div&gt;
&lt;/div&gt;


&lt;h2&gt;
  
  
  How I Built It
&lt;/h2&gt;

&lt;p&gt;SpudBot is built entirely in Python, utilizing local OS libraries (pyautogui, psutil, ctypes) to monitor screen time and manipulate the Windows environment.&lt;/p&gt;

&lt;p&gt;At its core, SpudBot relies on local open-weight models (like qwen3:8B) running locally via Ollama. Instead of relying on pre-written strings, the Python script feeds your current session duration and active window context to the local LLM, wrapped in a strict system prompt instructing it to act as an increasingly unhinged, sarcastic potato. This ensures every taunt is unique, contextual, and perfectly tailored to bully you into going outside.&lt;/p&gt;

&lt;h2&gt;
  
  
  Why Does Open Innovation Matter?
&lt;/h2&gt;

&lt;p&gt;For a project like this, open innovation isn't just a bonus; it’s a requirement.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;&lt;p&gt;Privacy: SpudBot constantly monitors active screen time, idle states, and potentially active window names. Sending this constant stream of personal desktop metadata to a closed-API server (like OpenAI or Anthropic) is a massive privacy risk. Open-source local inference guarantees my data never leaves my machine.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;Offline Capability: A true "Touch Grass" app shouldn't break when you lose Wi-Fi. SpudBot runs perfectly off-grid on a laptop.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;No Safety Filters on Sarcasm: Closed models often flag aggressive sarcasm or "mean" personas under their safety guardrails. Using an open-weight model allows me to fully tune the potato's unhinged, guilt-tripping personality without hitting corporate API filters.&lt;/p&gt;&lt;/li&gt;
&lt;li&gt;&lt;p&gt;Cost: Running a local background process that constantly generates text costs absolutely $0.&lt;/p&gt;&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Built in collaboration with &lt;a class="mentioned-user" href="https://dev.to/minioncodes"&gt;@minioncodes&lt;/a&gt; &lt;/p&gt;

</description>
      <category>devchallenge</category>
      <category>hf26challenge</category>
      <category>ai</category>
      <category>hacktoberfest</category>
    </item>
  </channel>
</rss>
